AMD On Track To Ship 32nm Chips In 2010

[Idontcare]

AMD On Track To Ship 32nm Chips In 2010

Dirk Meyer, chief executive of Advanced Micro Devices (NYSE: AMD), said Friday the chipmaker plans to "ramp up" production of next-generation 32-nanometer processors in the middle of next year with volume production starting in the fourth quarter.

http://www.informationweek.com...o=1&queryText=&isPrev=

Saw this over on XS, albeit that thread is lead-off with a far more optimistic viewpoint than the original InformationWeek article (linked above) lets on.

 

[allies]

Looks like good news... we'll see if it pans out.

I'm super interested in the 'dozer architecture, hearing as it's brand new tech from the ground up (well as much "from the ground up" as you can be nowadays)...

I smell an upgrade for me early 2011

 

[OCGuy]

Didnt they just get 45nm out?

Maybe they should work on performance before nm.

 

[imported_Asianman]

Late 2010 will be when bulldozer comes out. Hasn't AMD learned their lesson? they shouldn't change both a design and the manufacturing at once. If this flops, then AMD wouldn't look great. They should build the phenom 2s on the 32nm at the beginning of 2010, and then change the design at late 2010. They could, however, experiment by building 32nm notebook chips. That may be a great solution

 

[Cookie Monster]

Hmm, although bulldozer intrigues me abit (this could make or break AMD since they are taking some huge risks with their next gen architecture), westmere is taking all of my attention in the cpu department . Me wants one.

 

[Idontcare]

Originally posted by: Asianman
Late 2010 will be when bulldozer comes out. Hasn't AMD learned their lesson? they shouldn't change both a design and the manufacturing at once. If this flops, then AMD wouldn't look great. They should build the phenom 2s on the 32nm at the beginning of 2010, and then change the design at late 2010. They could, however, experiment by building 32nm notebook chips. That may be a great solution

It NOT might be a "what's the best case scenario" situation for AMD, they need product which commands higher ASP's and they are in no position to take the risk-conservative path here.

Building K10.5 on 32nm, or really doing anything but working on Bulldozer, requires AMD to allocate resources away from the bulldozer effort (lengthening its deliver timeline or reducing its capability so as to minimize delays)...AMD is not Intel, they can't just resource every tom, dick, and harry project idea just because it would be the best way to reduce risk. They have limited resources and a very real limitation on the time-horizon.

If they don't catch Intel with a competitive product before Intel transitions to 22nm then they are going to be on this perpetual treadmill of being one node behind and one architecture iteration behind, and that equals $100 ASP's which will not pay the bills heading into 16nm.

You notice AMD is intentionally staying out of the netbook market, they could field a netbook CPU but they simply cannot afford to undermine the existing Bulldozer program simply for the sake of creating a netbook project that would use up precious few resources AMD has at the moment.

 

[jones377]

Originally posted by: Idontcare

Originally posted by: Asianman
Late 2010 will be when bulldozer comes out. Hasn't AMD learned their lesson? they shouldn't change both a design and the manufacturing at once. If this flops, then AMD wouldn't look great. They should build the phenom 2s on the 32nm at the beginning of 2010, and then change the design at late 2010. They could, however, experiment by building 32nm notebook chips. That may be a great solution

It NOT might be a "what's the best case scenario" situation for AMD, they need product which commands higher ASP's and they are in no position to take the risk-conservative path here.

Building K10.5 on 32nm, or really doing anything but working on Bulldozer, requires AMD to allocate resources away from the bulldozer effort (lengthening its deliver timeline or reducing its capability so as to minimize delays)...AMD is not Intel, they can't just resource every tom, dick, and harry project idea just because it would be the best way to reduce risk. They have limited resources and a very real limitation on the time-horizon.

If they don't catch Intel with a competitive product before Intel transitions to 22nm then they are going to be on this perpetual treadmill of being one node behind and one architecture iteration behind, and that equals $100 ASP's which will not pay the bills heading into 16nm.

You notice AMD is intentionally staying out of the netbook market, they could field a netbook CPU but they simply cannot afford to undermine the existing Bulldozer program simply for the sake of creating a netbook project that would use up precious few resources AMD has at the moment.

But they are putting resources on other things than Bulldozer. It's just that all their other projects seem to fail one after the other. The most recent one being the 45nm version of Fusion.

 

[Rick James]

Originally posted by: Ocguy31
Didnt they just get 45nm out?

Maybe they should work on performance before nm.

X2. They're still getting slaughtered by Intel

 

[Idontcare]

Originally posted by: jones377
But they are putting resources on other things than Bulldozer.

Well of course they are, there will always be a ranking list of projects deemed critical to AMD's future that must be kept on life-support at all costs until the company ceases to exist.

Those projects carry with them an unacceptable risk of opportunity loss if the project is not pursued in some non-zero level of activity.

Shrinking K10.5 is a nice-to-have, not a need-to-have. It is a luxury item that AMD doesn't have the luxury of purchasing owing to the opportunity cost it would impart on other more critical projects, such as Bulldozer.

Originally posted by: jones377
It's just that all their other projects seem to fail one after the other. The most recent one being the 45nm version of Fusion.

That is a hallmark of being spread to thin, too few resources being allocated to too many projects. Exactly what they would be doing more of if they spread resources to cover both a K10.5 32nm shrink program and a 32nm Bulldozer program.

You don't trim your staff, reduce R&D expenditures, and keep the same number of high-resource consuming projects. Focusing across the board is an oxymoron.

An employee suggests setting priorities so they'll know how to apply their limited resources.

The manager's response: "Why can't we concentrate our resources across the board?"

http://www.ifsociety.org/voxma.../dilbert_principle.htm

 

[allies]

I guess people still don't realize that this is a delay of original projections. There is NOTHING that AMD can do right now to touch the i7 architecture. In the mid-lowhigh end market, they're performing well with the Phenom II. I think this is a good move, IMO.

 

[allies]

Originally posted by: jones377

Originally posted by: Idontcare

Originally posted by: Asianman
Late 2010 will be when bulldozer comes out. Hasn't AMD learned their lesson? they shouldn't change both a design and the manufacturing at once. If this flops, then AMD wouldn't look great. They should build the phenom 2s on the 32nm at the beginning of 2010, and then change the design at late 2010. They could, however, experiment by building 32nm notebook chips. That may be a great solution

It NOT might be a "what's the best case scenario" situation for AMD, they need product which commands higher ASP's and they are in no position to take the risk-conservative path here.

Building K10.5 on 32nm, or really doing anything but working on Bulldozer, requires AMD to allocate resources away from the bulldozer effort (lengthening its deliver timeline or reducing its capability so as to minimize delays)...AMD is not Intel, they can't just resource every tom, dick, and harry project idea just because it would be the best way to reduce risk. They have limited resources and a very real limitation on the time-horizon.

If they don't catch Intel with a competitive product before Intel transitions to 22nm then they are going to be on this perpetual treadmill of being one node behind and one architecture iteration behind, and that equals $100 ASP's which will not pay the bills heading into 16nm.

You notice AMD is intentionally staying out of the netbook market, they could field a netbook CPU but they simply cannot afford to undermine the existing Bulldozer program simply for the sake of creating a netbook project that would use up precious few resources AMD has at the moment.

But they are putting resources on other things than Bulldozer. It's just that all their other projects seem to fail one after the other. The most recent one being the 45nm version of Fusion.

It failed due to the complete lack of projected profitability.

 

[SlowSpyder]

Originally posted by: Asianman
Late 2010 will be when bulldozer comes out. Hasn't AMD learned their lesson? they shouldn't change both a design and the manufacturing at once. If this flops, then AMD wouldn't look great. They should build the phenom 2s on the 32nm at the beginning of 2010, and then change the design at late 2010. They could, however, experiment by building 32nm notebook chips. That may be a great solution

Along the lines of what IDC said, I think AMD is taking a gamble because, well what is their other choice? Bulldozer will either be a success and they'll hopefully live to fight another day, or if Bulldozer flops like the Phenom did that would probably be the end of AMD. They are taking a bit of a gamble and putting resources into making Bulldozer a success, I don't think they have another choice here.

Originally posted by: Ocguy31
Didnt they just get 45nm out?

Maybe they should work on performance before nm.

PhII may not be an i7 slayer, but at least it's competitive in the price range where lots of people buy. Before they were only competitve on the very low end, now they can atleast coexist with Intel in the middle ground and make a few bucks.

 

[exar333]

Originally posted by: SlowSpyder

Originally posted by: Asianman
Late 2010 will be when bulldozer comes out. Hasn't AMD learned their lesson? they shouldn't change both a design and the manufacturing at once. If this flops, then AMD wouldn't look great. They should build the phenom 2s on the 32nm at the beginning of 2010, and then change the design at late 2010. They could, however, experiment by building 32nm notebook chips. That may be a great solution

Along the lines of what IDC said, I think AMD is taking a gamble because, well what is their other choice? Bulldozer will either be a success and they'll hopefully live to fight another day, or if Bulldozer flops like the Phenom did that would probably be the end of AMD. They are taking a bit of a gamble and putting resources into making Bulldozer a success, I don't think they have another choice here.

Originally posted by: Ocguy31
Didnt they just get 45nm out?

Maybe they should work on performance before nm.

PhII may not be an i7 slayer, but at least it's competitive in the price range where lots of people buy. Before they were only competitve on the very low end, now they can atleast coexist with Intel in the middle ground and make a few bucks.

But they not really competing with Intel evenly in the "middle ground". Intel can still sell a high-end dual-core for $150-200 while AMD has to sell the same for $50-100. Quads are a little closer, but Intel is still making a nice profit off their $250-350 quads and they have made and continue to make uber-profits from the EE quads.

IDK is right, AMD needs to get competitive with Intel in performance AND pricing, or they are in serious trouble. This is definitely a "make or break" processor for AMD, and they likely have the most pressure they have ever felt on them right now to make this a win.

Edit: SP

 

[Glenn]

Hey, I'm all for AMD's success and make it a point to sell ATI cards right now with most all my builds but...

The CPU guys have been repeating this for too long now. Kinda like shooting at a duck 30yds down range and aiming right at it! You make alot of noise and everyone knows you took a shot but you missed the flyin duck because it's already 20 yards downrange from where you shot!

AMD's greatest successes came when they were leading the target. (Athlon, X64, IMC, etc...) I hope they survive, but it isn't looking too good right now.

 

[Nemesis 1]

I been trying to tell you guys that. The equipment AMD bought for 45nm . was already mostly setup to do 32nm. The big question was the chip designed for 32nm . SOI/Metal gates/ Amd problem isn't having equipent its the processor shrink . That may cause problems. Its like I7. Intel has said all along that I7 was designed for the 32nm process. Not the 45nm. At the time Intel was calling it Nehalem C. So much depends on how close is AMD to an actual chip designed for 32nm.

 

[Flipped Gazelle]

Originally posted by: ExarKun333
But they not really competing with Intel evenly in the "middle ground". Intel can still sell a high-end dual-core for $150-200 while AMD has to sell the same for $50-100. Quads are a little closer, but Intel is still making a nice profit off their $250-350 quads and they have made and continue to make uber-profits from the EE quads.

IDK is right, AMD needs to get competitive with Intel in performance AND pricing, or they are in serious trouble. This is definitely a "make or break" processor for AMD, and they likely have the most pressure they have ever felt on them right now to make this a win.

Edit: SP

AMD has a "high end dual core"? AMD competes with Intels "high end dual cores" with the Phenom II X3's, and they are priced $100-$150. AMD's PhII X4's compete very well with all but the highest of Intel's C2Q's. The big problem is that (most likely) AMD's making less $$ per chip than Intel.

 

[Flipped Gazelle]

Originally posted by: Nemesis 1
I been trying to tell you guys that. The equipment AMD bought for 45nm . was already mostly setup to do 32nm. The big question was the chip designed for 32nm . SOI/Metal gates/ Amd problem isn't having equipent its the processor shrink . That may cause problems. Its like I7. Intel has said all along that I7 was designed for the 32nm process. Not the 45nm. At the time Intel was calling it Nehalem C. So much depends on how close is AMD to an actual chip designed for 32nm.

Most of us who've been paying attention know this already.

Maybe AMD is hoping for a gov't bail-out...

 

[Martimus]

Originally posted by: Flipped Gazelle
Most of us who've been paying attention know this already.

Maybe AMD is hoping for a gov't bail-out...

They already got one, it just came from the UAE, and not the USA.

Also, Meyer stated that Bulldozer won't roll out until 2011, so if they are ramping production of 32nm in 2010, it will likely be the current architecture that goes first. Probably Magny-Cours, and the like.

 

[exar333]

Originally posted by: Flipped Gazelle

Originally posted by: ExarKun333
But they not really competing with Intel evenly in the "middle ground". Intel can still sell a high-end dual-core for $150-200 while AMD has to sell the same for $50-100. Quads are a little closer, but Intel is still making a nice profit off their $250-350 quads and they have made and continue to make uber-profits from the EE quads.

IDK is right, AMD needs to get competitive with Intel in performance AND pricing, or they are in serious trouble. This is definitely a "make or break" processor for AMD, and they likely have the most pressure they have ever felt on them right now to make this a win.

Edit: SP

AMD has a "high end dual core"? AMD competes with Intels "high end dual cores" with the Phenom II X3's, and they are priced $100-$150. AMD's PhII X4's compete very well with all but the highest of Intel's C2Q's. The big problem is that (most likely) AMD's making less $$ per chip than Intel.

AMD doesn't have a high-end dual core at all, they have been selling the X2's for <$100.00 for some time and they pale in comparison to any 5xxx, 7xxx, or 8xxx series. While the X3's are good chips, they are no E8500 that can hit 4.0ghz+ with ease. The X4's compete with Intel's lower to middle-ground Quads, but still sell for much less.

I don't think we disagree entirely, but keep in mind that you are referencing VERY new processors from AMD whereas there really has been no competition in the mid to upper range since C2D launched.

 

[Flipped Gazelle]

Originally posted by: ExarKun333

AMD doesn't have a high-end dual core at all, they have been selling the X2's for <$100.00 for some time and they pale in comparison to any 5xxx, 7xxx, or 8xxx series. While the X3's are good chips, they are no E8500 that can hit 4.0ghz+ with ease. The X4's compete with Intel's lower to middle-ground Quads, but still sell for much less.

I don't think we disagree entirely, but keep in mind that you are referencing VERY new processors from AMD whereas there really has been no competition in the mid to upper range since C2D launched.

That's what I mean: AMD has not had a "high-end dual core" in a couple of years, so it tickled me to see that phrase.

AMD's Kuma competes pretty well vs E5200, until you factor in overclocking.

X3's may not hit 4 Ghz with ease, but they regularly hit ~ 3.5, and have an additional core vs the E8x00. Overall, the X3 is a superior chip, IMO, but it really depends on the software the end user runs. If your main use is gaming, the E8x00 is the one to get, for now.

The vast majority of CPU's are never overclocked, so I am someone loathe to bring that into a discussion regarding a company's success.

What kills AMD is that Intel could easily slash prices and still make $$, whereas that's doubtful for AMD, to the best of my knowledge. If Intel started selling an E8500 for $100, or a Q9400 for $150, only the most ardent AMD fanboys wouldn't choose Intel.

 

[Idontcare]

Originally posted by: Flipped Gazelle
What kills AMD is that Intel could easily slash prices and still make $$, whereas that's doubtful for AMD, to the best of my knowledge.

Given that AMD doesn't make money now with its current prices it is a fairly safe bet to go out on a limb and state they to won't start to make money with lower prices.